The available evidence for treatment of coronavirus disease 2019 (COVID-19) is either indirect (from studies of influenza, severe acute respiratory syndrome and Middle East respiratory syndrome) or from several observational studies and randomized controlled trials in patients with COVID-19, which are limited in sample size and rigour, permitting only weak recommendations. • Given the inevitable adverse effects of interventions, the guideline panel (which included 2 patient partners) inferred that most informed patients would decline treatment when only very low-quality evidence of benefits — and, thus, very large uncertainty — is available. • The panel made only 1 weak recommendation in favour of treatment: use of corticosteroids in patients with acute respiratory distress syndrome (ARDS), based on indirect evidence. • The panel made weak recommendations against use of corticosteroids in patients without ARDS, against use of convalescent plasma and against several antiviral drugs that have been suggested as potential treatments for COVID-19. • Rigorous randomized trials are urgently needed to establish the benefits and risk of candidate interventions.